Layout documents

PDF to Word for jackets, spreads, and artwork

Illustrated pages fail in generic converters because the product is the layout, not the paragraph. A jacket has a title stacked on art; a spread has panels; a poster has type sitting on a painting. Rebuild Page treats those as layout jobs: reconstruct the page as editable HTML, compare it to the original, then build Word from that HTML.

What stays intact

Region order, image blocks, and the relationship between headlines and artwork. You will still review the reconstruction — illustrated pages are the hardest class — but you are not starting from a flattened text box. Edits in the studio are the source for the Word file.

When to use this instead of a letter conversion

Use the layout path when the page would look wrong as a Word outline: covers, dust jackets, illustrated book interiors, marketing one-pagers, and bilingual posters. Running text letters can use the lighter letter path. Card-shaped scans belong on ID cards to Word.

How a layout job runs

1. Upload the PDF or page images in the workspace.
2. Auto-detect should choose layout for illustrated spreads. API callers can set document_type to layout.
3. Open the side-by-side studio. Fix type on top of art if OCR missed a line.
4. Download Word generated from the HTML you approved.

Step-by-step for every document class is on how it works. Products that need the same output can call the Partner API.

Credits

Artwork pages hold more credits per page than a plain letter, then settle to what the conversion actually used. Prepaid packs, no subscription — see pricing. Uploads are not used to train models.

Questions

What counts as an illustrated PDF?

Book jackets, magazine or comic spreads, posters, and pages where artwork and text share the same canvas — not a simple one-column letter.

Why do other converters flatten jackets?

They emit Word from a text layer and ignore region geometry. Rebuild Page rebuilds the spread as HTML first, so panels, titles, and images keep their places when Word is exported.

Can I edit text on the reconstructed spread?

Yes. The studio is the HTML reconstruction. Change copy or fix OCR, then download Word from that HTML so the file matches what you approved.

How do I call this from an API?

Use document_type layout (or auto) on POST /v1/jobs, then GET the job’s DOCX. Same pipeline as the studio. See the Partner API docs.